Suspicion of recurrence should prompt evaluation with imaging to assess anatomy, extent of localized and/or disseminated disease, and to determine resectability. Differentiating between tumor and inflammatory changes can be difficult and a complement of imaging modalities is often necessary [16]. Computed tomography (CT) of the chest, abdomen, and pelvis should be obtained to assess for extrapelvic metastasis and the extent of local involvement. Since CT is less discerning for predicting local tumor infiltration and adjacent organ involvement, magnetic resonance imaging (MRI) of the pelvis should typically be performed to evaluate for invasion of surrounding structures [17, 18]. High resolution MRI has been shown to have a negative predictive value of 93–100 % for identification of local invasion, however is also limited in its ability to distinguish recurrence from diffuse fibrosis [19–21]. Despite lower accuracy for anterior recurrence close to the bladder and mucinous tumors, PET – CT has a sensitivity and specificity of nearly 100 and 96 %, respectively, for diagnosing local recurrence [22–28].

For any patient with a suspicion for recurrence, histologic proof should be vigorously sought prior to proceeding with surgery, though at times can be difficult to definitively obtain. If unable to confirm histologically, other factors should be taken into consideration such as an increase in the dimensions of the area of interest over time, invasion of surrounding structures, a rising carcinoembryonic antigen (CEA), the development of symptoms (e.g. pain from neural or osseous involvement, urinary, fecal, or neurologic complications, bleeding), and overall multidisciplinary assessment [29].

Different classification systems to assess tumor resectability and provide prognostic information are based on the anatomic location of pelvic recurrence, degree and site(s) of fixation, and symptoms. None can unfailingly predict resectability prior to surgery since new findings may be discovered intraoperatively. The system utilized at the Mayo Clinic classifies recurrence based on pain (S0 – asymptomatic, S1 – symptomatic without pain, S2 – symptomatic with pain) and fixation to surrounding structures (F0 – not fixed, F1 – fixed to 1 site, F2 – fixed to 2 sites, F3 – fixed to ≥3 sites). Fixation is defined anatomically as anterior, posterior, and lateral [30, 31]. Wanebo developed a system identifying five stages of invasion. (TR1) limited muscularis propria invasion, (TR2) full thickness muscularis propria involvement, (TR3) anastomotic recurrence into perirectal soft tissue, (TR4) adjacent organ invasion/not fixed, and (TR5) invasion of sidewalls or bony ligaments [32]. A system proposed by the Leeds group classifies recurrence as central (confined to pelvic organs without osseous involvement), sidewall (involving lateral sidewall), sacral (abutting or invading sacrum), or composite (sacral and sidewall involvement) [33]. Yamada and colleagues devised a system evaluating the pattern of pelvic fixation as localized (adjacent organs or tissue), sacral (lower sacrum – S3/S4/S5, coccyx, periosteum), and lateral (sciatic nerve, greater sciatic foramen, lateral sidewall, upper sacrum – S1/S2) [34]. Memorial Sloan Kettering utilizes a system based on involvement of surrounding structures and anatomic location; axial (anastomotic recurrence, perineal and perirectal invasion), anterior (urogenital involvement), posterior (presacral fascia or sacral invasion), and lateral (sidewall and bony pelvis involvement) [35]. The Royal Marsden Hospital system evaluates the extent of tumor invasion in seven different pelvic compartments based on MRI; (C) central, (P) posterior, (I) inferior, (L) lateral, (PR) peritoneal reflection, (AA-PR) anterior above and (AB-PR) anterior below the peritoneal reflection [36].

Hruby and colleagues evaluated sites of pelvic recurrence in 269 patients with rectal cancer untreated with radiotherapy and found nearly 90 % in the posterior or central pelvis, with 20 % at the level of the anastomosis [32]. The Mayo system found worse outcomes in those presenting with pain and increased points of fixation [30, 31]. Yamada found a 5 year survival rate of 38 % for localized disease, 10 % for sacral involvement, and 0 % for lateral invasion [34]. The Memorial Sloan Kettering group documented the likelihood of a R0 resection for axial only recurrence as 90 %, versus 36 % for lateral involvement [24]. Based on findings from the Royal Marsden Hospital, survival is decreased when MRI reveals involvement of more than two compartments, or when the lateral or posterior planes are involved [36].

Surgery, chemotherapy, and radiation alone result in high rates of local and distant failure; but when combined in a multimodal fashion have been shown to improve local control, survival, rates of salvage surgery, and resection with R0 margins [7, 31, 38–44]. Radiotherapy naïve patients should receive a full course (5040 cGy/50.4 Gy) of external beam radiotherapy (EBRT) administered concurrently with sensitizing 5-floururacil (5-FU) based chemotherapy The addition of other cytotoxic (oxaliplatin, irinotecan) and biologic (cetuximab, bevacizumab) agents along with 5-FU has not shown benefit to date [45]. For those previously irradiated, a hyperfractionated course of 2000–3000 cGy EBRT along with 5-FU can be completed prior to surgery. Although safe and effective, there is a lack of high quality data to support this approach [39, 42, 46, 47]. Intensity modulated radiotherapy reduces the dose of radiation to surrounding structures, though supporting evidence is limited regarding a benefit over conventional radiotherapy [48]. To maximize tumor response, surgery is planned for 6–8 weeks following completion of radiation [49].

As part of a multimodal treatment approach, intraoperative radiotherapy (IORT) has been shown to increase survival by 15 % or more and improve local control in selected patients [50, 51]. Intraoperative radiotherapy overcomes the dose restriction of EBRT by limiting exposure to surrounding unaffected structures. The total dose administered is dependent on preceding amounts of preoperative radiotherapy delivered. The Mayo Clinic has a dedicated operating room with a linear accelerator to provide electron beam radiotherapy, and a dose of 1000 cGy is given for minimal residual disease (margin microscopically involved or clear by <5 mm), 1500 cGy for unresectable gross disease less than 2 cm, and 2000 cGy for more than 2 cm [52]. Other means of administering locally directed radiation include high dose intraoperative brachytherapy (HDR-IORT), perioperative brachytherapy, and photon radiosurgery [53, 54]. Multiple institutions have shown improved disease free and overall survival, as well as local control, following R0 and microscopically positive (R1) resections when IORT is incorporated into a multimodal treatment regimen, regardless of the IORT approach chosen [50, 51, 55–62]. Others have not been able to document a benefit [42, 63–65]. The largest series evaluating 304 patients with locally recurrent rectal cancer was reported by the Mayo Clinic in 2003. Of those, 138 underwent a R0 resection, 27 a R1, and 139 had gross (R2) residual disease. The 5 year survival rates were greatest for R0 versus the R1 and R2 resections (37 versus 16 %, p < 0.001). Survival after extended procedures (sacrectomy, pelvic exenteration, cystectomy with ileal conduit) was comparable to more limited resections (28 versus 21 %, p = 0.11) [31]. Overall, results from specialized centers support an oncologic advantage for IORT in select patients. However, there is a significant amount of heterogeneity between centers, making broad consensus statements challenging.

A R0 resection provides the highest rate of local control as well as cancer specific and overall survival. The presence of microscopically or grossly positive (R2) margins decreases survival [66]. Resection is based on defining invasion into adjacent structures, as well as the presence of metastatic disease. Factors typically associated with the inability to pursue a curative (R0) resection include poor performance status, encasement of external iliac vessels, presence of venous or lymphatic obstruction, distant metastasis, fixation to two or more sites (F2 or F3 involvement), predicted R1/R2 resection, sacral invasion above S2, extension though the greater sciatic notch, circumferential or multiple sites of pelvic sidewall involvement, bilateral ureteral obstruction outside the bladder trigone, and S1/S2 nerve root involvement [35, 67–70].

Upwards of 50 % of patients with local recurrence will have a metastatic lesion noted during initial evaluation. If resectable, surgical intervention can be pursued in a synchronous or staged approach, and in highly selected patients outcomes are favorable [71, 72]. En bloc resection of involved ureteral or iliac vessels is possible and is associated with an increased R0 resection rate [6, 35, 69, 73–75]. Extended resections to achieve negative margins, including a high sacrectomy, improve local control and survival [6, 8, 68, 73, 76–84], though can result in significant life altering morbidity. Complications are higher for fixed tumors, and reduced tumor free resection margins and survival has been demonstrated in those with symptomatic pain and fixation to more than one area [30, 31]. Other factors noted to decrease R0 resections are male sex, increased age, previous abdominoperineal resection, higher stage of primary tumor, and elevated CEA level [35]. If the morbidity of an extended resection to obtain a R0 margin outweighs the potential benefit, an R1 resection with IORT should be pursued.

Following radiation, surgery, and IORT, complications occur in upwards of 65 % of patients owing to the heavily irradiated field [31, 46]. The most common complications include pelvic abscesses (6.6 %), bowel obstruction (5.3 %), enteric fistulas (4.3 %), and wound complications (4.6 %). Those with extended resections and more than two sites of fixed recurrence, experience the highest rates of postoperative complications [31]. Nelson and colleagues from the Mayo Clinic reported reduced wound complications and length of hospital stay when flap repairs were utilized in comparison to primary closure. Other studies have corroborated the benefit of perineal defect closure/reconstruction with techniques including omentoplasty combined with biologic implants, vertical rectus or myocutaneous oblique abdominis muscle flaps, gluteal rotation flaps, gracilis flaps, and free flaps [85–91].

If cure is not possible (R2 resection would be required), then palliation of symptoms may be sought with a combination of modalities (chemoradiotherapy, urinary and colonic stents, nephrostomy tubes, endoscopic laser ablation, targeted surgery), all of which have been shown to improve quality of life, though rarely halt disease progression [92–96]. External beam radiation, therapy (EBRT) including reirradiation when necessary, has been noted to control pain in 50–90 % of patients [97, 98]. The addition of chemotherapy to EBRT also improves symptoms, but not 5 year survival [41, 99, 100]. Improvement in symptoms and quality of life has been shown to be superior following surgery compared to non-surgical approaches, even for selective cases with distant metastasis, although less so for extended resections [25, 101, 102].